---
title: "Copenhagen — research institutions and their papers"
description: "Copenhagen (Denmark): 5 institutions with papers on Elamur Research — 4 universities, 1 hospital — and 16,112 citations of their works."
canonical: "https://research.elamur.ai/free/en/agglomerations/copenhagen/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/agglomerations/copenhagen/)

[home](https://md.research.elamur.ai/free/en/) / [institutions](https://md.research.elamur.ai/free/en/orgs/) / [agglomerations](https://md.research.elamur.ai/free/en/agglomerations/) / Copenhagen

# Copenhagen

Agglomeration [Denmark](https://md.research.elamur.ai/free/en/countries/denmark/) · metro population 1,987,482 (functional urban area, GHSL 2015)

**5** institutions with a page**76** works in our library**16,112** citations of their works

## About

### Profile

Copenhagen is a city region in [Denmark](https://md.research.elamur.ai/free/en/countries/denmark/). On this site its research is carried by four universities and one hospital: the [University of Copenhagen](https://md.research.elamur.ai/free/en/universities/university-of-copenhagen/), the [IT University of Copenhagen](https://md.research.elamur.ai/free/en/universities/it-university-of-copenhagen/), [Copenhagen Business School](https://md.research.elamur.ai/free/en/universities/copenhagen-business-school/), the [Technical University of Denmark](https://md.research.elamur.ai/free/en/universities/technical-university-of-denmark/) in Kongens Lyngby, and [Copenhagen University Hospital](https://md.research.elamur.ai/free/en/hospitals/copenhagen-university-hospital/).

The works collected here from the region come almost entirely from one field, natural language processing and computational linguistics. They include studies of how hateful or offensive language can be found automatically, of what large pretrained language models learn and how they can be adapted, of tools for languages with few resources, of the social and ethical side of language technology, and of linguistic phenomena such as ellipsis. A smaller medical strand applies text mining and large language models to clinical writing.

The University of Copenhagen accounts for by far the largest share of the region's works and citations on this site. The IT University of Copenhagen is the second centre. Copenhagen Business School, the Technical University of Denmark and Copenhagen University Hospital each contribute a few works, and these are often tied to one of the two larger centres through shared papers.

### History

Among the works here, the earliest is [Generation of VP Ellipsis: A Corpus-Based Approach](https://md.research.elamur.ai/free/en/papers/hardt-2001-generation-vp-ellipsis-corpus-based/) from 2001, associated with Copenhagen Business School. It asks when a language generation system ought to drop a verb phrase and keep only an auxiliary, and it treats this as a choice that can be learned from a corpus. The interest in ellipsis returns in later work from the same school, in 2016 and 2021, making it the region's longest continuous line of research on this site.

The next dated work is from 2014. It studies negation scope and spelling variation in Danish electronic patient records and is shared between the Technical University of Denmark and Copenhagen University Hospital. It describes a matcher that uses a dictionary of diagnostic codes on psychiatric records, and it opens the region's clinical text-mining strand among the works here.

The volume of work grows sharply in 2016. In that year the University of Copenhagen produced several of the most cited papers associated with the region: two studies of hate speech on Twitter, a position paper on the social impact of language processing, and a paper on deep multi-task learning. From 2017 to 2021 the output widens to emoji-based representation learning, the limits of unsupervised bilingual dictionary induction, parallel corpora for low-resource languages, a survey of BERT, a survey of domain adaptation, a multi-task learning toolkit, Danish offensive-language detection and debates about evaluation.

The most recent works here are from 2023 and 2024. In 2023 Copenhagen University Hospital took part in a study of large language models applied to summarising clinical text. In 2024 the IT University of Copenhagen took part in a multilingual instruction-tuning collection. Together they show the region's research moving toward large language models in both general and medical settings.

### Research profile

The clearest strength of the region on this site is the study of abusive and harmful language online. [Hateful Symbols or Hateful People? Predictive Features for Hate Speech Detection](https://md.research.elamur.ai/free/en/papers/waseem-2016-hateful-symbols-or-hateful-people/), from the University of Copenhagen, builds classifiers that flag tweets carrying racism or sexism, and it asks which features best predict such content. A companion paper, [Are You a Racist or Am I Seeing Things? Annotator Influence on Hate Speech Detection](https://md.research.elamur.ai/free/en/papers/waseem-2016-are-you-racist-or-am/), turns to the labels themselves. It examines whether the expertise of the people who annotate hateful posts changes the quality of the resulting classifiers, comparing crowd workers with more expert annotators. At the IT University of Copenhagen, [Offensive Language and Hate Speech Detection for Danish](https://md.research.elamur.ai/free/en/papers/sigurbergsson-2020-offensive-language-hate-speech-detection/) carries the same concern into Danish; its authors present it as the first collection of its kind for that language.

A second, closely related line concerns misinformation and evaluation. [SemEval-2019 Task 7: RumourEval](https://md.research.elamur.ai/free/en/papers/gorrell-2019-semeval-2019-task-7-rumoureval/), linked to the IT University of Copenhagen, set out a shared task and benchmark for judging whether rumours are true and how social media users back or deny them. [We Need to Consider Disagreement in Evaluation](https://md.research.elamur.ai/free/en/papers/basile-2021-we-need-consider-disagreement-evaluation/) argues against scoring systems only against one merged gold label. It holds that genuine disagreement among annotators should be part of how language systems are judged, which connects back to the earlier work on annotator influence in hate speech.

A third strand looks at the ethics and social consequences of language technology. [The Social Impact of Natural Language Processing](https://md.research.elamur.ai/free/en/papers/hovy-2016-social-impact-natural-language-processing/), from the University of Copenhagen, borrows terms from the ethics of technology to sort the harms that language processing can cause, among them the exclusion of some groups of users.

A fourth strand concerns representation learning and the inner workings of neural models. [A Primer in BERTology: What We Know About How BERT Works](https://md.research.elamur.ai/free/en/papers/rogers-2020-primer-bertology-what-we-know/) surveys what research had found about the BERT model. It reviews what the model appears to encode and covers suggested modifications to how it is trained, how it is built and how it is fine-tuned. [Using millions of emoji occurrences to learn any-domain representations](https://md.research.elamur.ai/free/en/papers/felbo-2017-millions-emoji-occurrences-learn-any/) starts from the scarcity of annotated examples for recognising feelings, moods and irony in text. It uses emoji in social media posts as a large source of distant supervision for learning representations that transfer to those tasks.

Multi-task and transfer learning form a fifth strand. [Deep multi-task learning with low level tasks supervised at lower layers](https://md.research.elamur.ai/free/en/papers/sogaard-2016-deep-multi-task-learning-low/) shows that simpler linguistic tasks can be supervised at lower layers of a deep network and higher-level tasks at upper layers, offering a way to join cascaded pipelines with shared representations. At the IT University of Copenhagen, [Massive Choice, Ample Tasks (MaChAmp)](https://md.research.elamur.ai/free/en/papers/goot-2021-massive-choice-ample-tasks-machamp/) provides a toolkit for multi-task learning, in which users set out their corpora and tasks in a configuration file. [Neural Unsupervised Domain Adaptation in NLP–-A Survey](https://md.research.elamur.ai/free/en/papers/ramponi-2020-neural-unsupervised-domain-adaptation-nlp/) reviews ways to adapt models to new kinds of text without labels in the target domain. It proposes the wider notion of a language "variety", drawing on Plank's variety space, in place of the narrower idea of a domain.

A sixth strand serves languages with few resources and cross-lingual methods. [JW300: A Wide-Coverage Parallel Corpus for Low-Resource Languages](https://md.research.elamur.ai/free/en/papers/agic-2019-jw300-wide-coverage-parallel-corpus/) responds to the shortage of translated text for many languages with a broad parallel corpus. [Cross-Lingual Parser Selection for Low-Resource Languages](https://md.research.elamur.ai/free/en/papers/agic-2017-cross-lingual-parser-selection-low/) proposes a lightweight way to choose, while the system runs, which source-language parser to transfer to a target language. [On the Limitations of Unsupervised Bilingual Dictionary Induction](https://md.research.elamur.ai/free/en/papers/sogaard-2018-limitations-unsupervised-bilingual-dictionary-induction/) tests the assumption behind adversarial methods for building bilingual dictionaries without supervision and examines where such methods fail. The 2024 [Aya Dataset](https://md.research.elamur.ai/free/en/papers/singh-2024-aya-dataset-open-access-collection/) extends this multilingual interest to instruction tuning. It starts from the observation that curated prompt and completion pairs are overwhelmingly English and assembles an open collection across many languages.

A seventh, smaller strand is theoretical and computational work on ellipsis, centred on Copenhagen Business School. Besides the 2001 generation paper, [Antecedent Selection for Sluicing: Structure and Content](https://md.research.elamur.ai/free/en/papers/anand-2016-antecedent-selection-sluicing-structure-content/) holds that finding the antecedent of a sluice depends on syntax as well as discourse, and it designs features describing how the two are related in structure. [Ellipsis Resolution as Question Answering: An Evaluation](https://md.research.elamur.ai/free/en/papers/aralikatte-2021-ellipsis-resolution-as-question-answering/) trains models built on BERT for sluicing and verb-phrase ellipsis, each on its own at first and later together with question-answering tasks.

Finally, the clinical strand links language technology to medicine. The 2014 study on [negation scope and spelling variation](https://md.research.elamur.ai/free/en/papers/thomas-2014-negation-scope-spelling-variation-text/) in Danish psychiatric records deals with the practical difficulties of mining clinical notes. [Adapted Large Language Models Can Outperform Medical Experts in Clinical Text Summarization](https://md.research.elamur.ai/free/en/papers/veen-2023-adapted-large-language-models-outperform/) starts from the time clinicians spend turning health-record content into reports, and it tests whether adapted language models can take on that work.

### Leading institutions and units

The University of Copenhagen is the region's main producer of research on this site. Its works cover hate speech detection, the social impact of language processing, multi-task learning, emoji-based representations, bilingual dictionary induction, cross-lingual corpora and the analysis of BERT. Through shared works it is the hub that most of the region's other institutions connect to.

The IT University of Copenhagen is the second centre. Its works in our library deal with low-resource and cross-lingual processing, domain adaptation, multi-task tooling, rumour verification, Danish offensive language, annotator disagreement and multilingual instruction tuning. Its profile leans toward resources, benchmarks and tools.

Copenhagen Business School contributes the ellipsis line, from the 2001 generation study to the 2021 evaluation of ellipsis resolution as question answering. The Technical University of Denmark in Kongens Lyngby appears through the emoji representation paper and the Danish patient-record study. Copenhagen University Hospital appears through the patient-record study and the 2023 work on clinical summarisation with large language models.

### Leading scientists

At the University of Copenhagen, [Dirk Hovy](https://md.research.elamur.ai/free/en/authors/dirk-hovy/) comes first. Hovy's work in our library includes the position paper on the social impact of language processing, which brought vocabulary from technology ethics into the discussion of harms. [Anders Søgaard](https://md.research.elamur.ai/free/en/authors/anders-sogaard/) follows. Søgaard is first author of the 2016 study on supervising low-level tasks at lower network layers and of the 2018 study on the limits of unsupervised bilingual dictionary induction, and is also associated with Copenhagen Business School.

[Isabelle Augenstein](https://md.research.elamur.ai/free/en/authors/isabelle-augenstein/) is the next name at the university. [Anna Rogers](https://md.research.elamur.ai/free/en/authors/anna-rogers/), who appears at both the University of Copenhagen and the IT University of Copenhagen, is first author of the BERTology primer, one of the most cited works associated with the region. The university's people also include [Ilias Chalkidis](https://md.research.elamur.ai/free/en/authors/ilias-chalkidis/), [Desmond Elliott](https://md.research.elamur.ai/free/en/authors/desmond-elliott/), [Daniel Hershcovich](https://md.research.elamur.ai/free/en/authors/daniel-hershcovich/) and [Christina Lioma](https://md.research.elamur.ai/free/en/authors/christina-lioma/). Lioma is also associated with the Technical University of Denmark.

At the IT University of Copenhagen, [Barbara Plank](https://md.research.elamur.ai/free/en/authors/barbara-plank/) comes first. Plank's notion of a variety space underpins the 2020 survey of neural unsupervised domain adaptation. [Leon Derczynski](https://md.research.elamur.ai/free/en/authors/leon-derczynski/) follows. [Željko Agić](https://md.research.elamur.ai/free/en/authors/zeljko-agic/) is first author of the JW300 parallel corpus and of the cross-lingual parser selection paper, both aimed at languages with few resources. [Rob van der Goot](https://md.research.elamur.ai/free/en/authors/rob-van-der-goot/) is first author of the MaChAmp multi-task toolkit. The university's people also include [Mike Zhang](https://md.research.elamur.ai/free/en/authors/mike-zhang/), [Natalie Schluter](https://md.research.elamur.ai/free/en/authors/natalie-schluter/) and [Elisa Bassignana](https://md.research.elamur.ai/free/en/authors/elisa-bassignana/).

At Copenhagen Business School, [Daniel Hardt](https://md.research.elamur.ai/free/en/authors/daniel-hardt/) is first author of the 2001 paper on generating verb-phrase ellipsis, the earliest work here and the start of the school's ellipsis research. [Nizar Habash](https://md.research.elamur.ai/free/en/authors/nizar-habash/) and [Fei Liu](https://md.research.elamur.ai/free/en/authors/fei-liu/) also appear in the school's circle on this site.

The names associated with the Technical University of Denmark include [Fan Yang](https://md.research.elamur.ai/free/en/authors/fan-yang/), [Yifan Du](https://md.research.elamur.ai/free/en/authors/yifan-du/), [Qi Zhang](https://md.research.elamur.ai/free/en/authors/qi-zhang/), [Stella Frank](https://md.research.elamur.ai/free/en/authors/stella-frank/), [Gerhard Weikum](https://md.research.elamur.ai/free/en/authors/gerhard-weikum/), [Zhenghao Liu](https://md.research.elamur.ai/free/en/authors/zhenghao-liu/) and [Cyril Goutte](https://md.research.elamur.ai/free/en/authors/cyril-goutte/), alongside Christina Lioma. At Copenhagen University Hospital, [Anuj Pareek](https://md.research.elamur.ai/free/en/authors/anuj-pareek/) is the person with a profile here, linked to the hospital's clinical research with language models.

### Landmark works

The most cited work associated with the region on this site is the BERTology primer from 2020. It brought together scattered findings about what BERT represents and how it might be improved, collecting them into a single overview of the model.

Close behind is the 2016 paper on predictive features for hate speech detection, which frames hateful tweets as a supervised classification problem and asks which kinds of features carry the signal. Its companion study on annotator influence is also among the most cited works here. It draws attention to the effect of who does the labelling, a question that the 2021 paper on disagreement in evaluation develops further.

The 2016 position paper on the social impact of language processing is another of the most cited works, offering a structured account of the harms language systems can cause. The 2017 emoji paper is among the most cited as well; it shows that the emoji people use can supervise models for affect-related tasks across domains.

Other much-cited works include the JW300 corpus, which widened the parallel text available for many low-resource languages, the multilingual Aya collection of 2024, and the 2018 study that questioned how robust unsupervised bilingual dictionary induction really is. The 2016 multi-task learning paper, the 2021 disagreement paper, the domain adaptation survey and the MaChAmp toolkit complete the region's most cited works in our library.

### Collaborations

The region's institutions are tied together by shared works. The JW300 corpus is associated with both the University of Copenhagen and the IT University of Copenhagen, and Anna Rogers appears at both. The emoji representation paper links the University of Copenhagen with the Technical University of Denmark, and Christina Lioma is associated with both. Anders Søgaard connects the University of Copenhagen with Copenhagen Business School.

The clearest link between academia and clinical practice is the 2014 study of Danish psychiatric patient records, shared between the Technical University of Denmark and Copenhagen University Hospital, which applies text-mining methods to hospital documents. The hospital's 2023 work on clinical summarisation continues this meeting of language technology and medicine. Several of the region's works are also community efforts, such as the RumourEval shared task and the multilingual Aya collection.

### Timeline

The earliest period on this site is marked by a single work. In 2001 Copenhagen Business School published the corpus-based study of when to generate verb-phrase ellipsis, starting the school's long interest in elliptical constructions.

In 2014 the clinical strand appears, with a study shared by the Technical University of Denmark and Copenhagen University Hospital on negation and spelling variation in Danish psychiatric records.

The year 2016 is a turning point. The University of Copenhagen produced both hate speech papers, the position paper on social impact and the study of multi-task learning across network layers. In the same year Copenhagen Business School published its work on antecedent selection for sluicing.

From 2017 to 2019 the range widens. The emoji representation paper and the cross-lingual parser selection study appeared in 2017, the bilingual dictionary induction critique followed in 2018, and the JW300 corpus and the RumourEval shared task came in 2019.

The years 2020 and 2021 bring surveys, tools and position papers. The BERTology primer, the domain adaptation survey and the Danish offensive-language collection appeared in 2020. The MaChAmp toolkit, the call to bring annotator disagreement into evaluation, and the evaluation of ellipsis resolution as question answering followed in 2021.

The latest period turns toward large language models. In 2023 Copenhagen University Hospital took part in showing that adapted models can rival experts at summarising clinical text. In 2024 the IT University of Copenhagen contributed to the Aya collection for multilingual instruction tuning.

## Institutions (5)

4 universities, 1 hospital.

### 🎓 Universities (4)

1. **[University of Copenhagen](https://md.research.elamur.ai/free/en/universities/university-of-copenhagen/)**University · Copenhagen, [Denmark](https://md.research.elamur.ai/free/en/countries/denmark/)*65 works · 12,760 citations*
2. **[IT University of Copenhagen](https://md.research.elamur.ai/free/en/universities/it-university-of-copenhagen/)**University · Copenhagen, [Denmark](https://md.research.elamur.ai/free/en/countries/denmark/)*11 works · 1,744 citations*
3. **[Copenhagen Business School](https://md.research.elamur.ai/free/en/universities/copenhagen-business-school/)**University · Copenhagen, [Denmark](https://md.research.elamur.ai/free/en/countries/denmark/)*3 works · 48 citations*
4. **[Technical University of Denmark](https://md.research.elamur.ai/free/en/universities/technical-university-of-denmark/)**University · Kongens Lyngby, [Denmark](https://md.research.elamur.ai/free/en/countries/denmark/)*2 works · 626 citations*

### 🏥 Hospitals and medical centres (1)

1. **[Copenhagen University Hospital](https://md.research.elamur.ai/free/en/hospitals/copenhagen-university-hospital/)**Hospital · Copenhagen, [Denmark](https://md.research.elamur.ai/free/en/countries/denmark/)*2 works · 934 citations*

## Works in our library (76)

Works of its institutions, the most cited first.

1. [**Hateful Symbols or Hateful People? Predictive Features for Hate Speech Detection on Twitter**Zeerak Waseem, Dirk Hovy*2016 · 2,158 citations*](https://md.research.elamur.ai/free/en/papers/waseem-2016-hateful-symbols-or-hateful-people/)
2. [**A Primer in BERTology: What We Know About How BERT Works**Anna Rogers, Olga Kovaleva, Anna Rumshisky*2020 · 2,127 citations*](https://md.research.elamur.ai/free/en/papers/rogers-2020-primer-bertology-what-we-know/)
3. [**Adapted Large Language Models Can Outperform Medical Experts in Clinical Text Summarization**Dave Van Veen, Cara Van Uden, Louis Blankemeier, et al.*2023 · 925 citations*](https://md.research.elamur.ai/free/en/papers/veen-2023-adapted-large-language-models-outperform/)
4. [**Are You a Racist or Am I Seeing Things? Annotator Influence on Hate Speech Detection on Twitter**Zeerak Waseem*2016 · 760 citations*](https://md.research.elamur.ai/free/en/papers/waseem-2016-are-you-racist-or-am/)
5. [**Using millions of emoji occurrences to learn any-domain representations for detecting sentiment, emotion and sarcasm**Bjarke Felbo, Alan Mislove, Anders Søgaard, et al.*2017 · 617 citations*](https://md.research.elamur.ai/free/en/papers/felbo-2017-millions-emoji-occurrences-learn-any/)
6. [**Deep multi-task learning with low level tasks supervised at lower layers**Anders Søgaard, Yoav Goldberg*2016 · 584 citations*](https://md.research.elamur.ai/free/en/papers/sogaard-2016-deep-multi-task-learning-low/)
7. [**The Social Impact of Natural Language Processing**Dirk Hovy, Shannon L. Spruit*2016 · 516 citations*](https://md.research.elamur.ai/free/en/papers/hovy-2016-social-impact-natural-language-processing/)
8. [**On the Limitations of Unsupervised Bilingual Dictionary Induction**Anders Søgaard, Sebastian Ruder, Ivan Vulić*2018 · 396 citations*](https://md.research.elamur.ai/free/en/papers/sogaard-2018-limitations-unsupervised-bilingual-dictionary-induction/)
9. [**JW300: A Wide-Coverage Parallel Corpus for Low-Resource Languages**Željko Agić, Ivan Vulić*2019 · 373 citations*](https://md.research.elamur.ai/free/en/papers/agic-2019-jw300-wide-coverage-parallel-corpus/)
10. [**Neural Unsupervised Domain Adaptation in NLP–-A Survey**Alan Ramponi, Barbara Plank*2020 · 334 citations*](https://md.research.elamur.ai/free/en/papers/ramponi-2020-neural-unsupervised-domain-adaptation-nlp/)
11. [**MultiFC: A Real-World Multi-Domain Dataset for Evidence-Based Fact Checking of Claims**Isabelle Augenstein, Christina Lioma, Dongsheng Wang, et al.*2019 · 318 citations*](https://md.research.elamur.ai/free/en/papers/augenstein-2019-multifc-real-world-multi-domain/)
12. [**Identifying beneficial task relations for multi-task learning in deep neural networks**Joachim Bingel, Anders Søgaard*2017 · 305 citations*](https://md.research.elamur.ai/free/en/papers/bingel-2017-identifying-beneficial-task-relations-multi/)
13. [**SemEval-2019 Task 7: RumourEval, Determining Rumour Veracity and Support for Rumours**Genevieve Gorrell, Elena Kochkina, Maria Liakata, et al.*2019 · 290 citations*](https://md.research.elamur.ai/free/en/papers/gorrell-2019-semeval-2019-task-7-rumoureval/)
14. [**Challenges and Strategies in Cross-Cultural NLP**Daniel Hershcovich, Stella Frank, Heather Lent, et al.*2022 · 290 citations*](https://md.research.elamur.ai/free/en/papers/hershcovich-2022-challenges-strategies-cross-cultural-nlp/)
15. [**Assessing Cross-Cultural Alignment between ChatGPT and Human Societies: An Empirical Study**Yong Cao, Li Zhou, Seolhwa Lee, et al.*2023 · 278 citations*](https://md.research.elamur.ai/free/en/papers/cao-2023-assessing-cross-cultural-alignment-between/)
16. [**A Diagnostic Study of Explainability Techniques for Text Classification**Pepa Atanasova, Jakob Grue Simonsen, Christina Lioma, et al.*2020 · 273 citations*](https://md.research.elamur.ai/free/en/papers/atanasova-2020-diagnostic-study-explainability-techniques-text/)
17. [**LexGLUE: A Benchmark Dataset for Legal Language Understanding in English**Ilias Chalkidis; Abhik Jana; Dirk Hartung; Michael Bommarito; Ion Androutsopoulos; Daniel Martin Katz;, Nikolaos Aletras*2021 · 256 citations*](https://md.research.elamur.ai/free/en/papers/chalkidis-2021-lexglue-benchmark-dataset-legal-language/)
18. [**Factuality Challenges in the Era of Large Language Models**Isabelle Augenstein, Timothy Baldwin, Meeyoung Cha, et al.*2023 · 241 citations*](https://md.research.elamur.ai/free/en/papers/augenstein-2023-factuality-challenges-era-large-language/)
19. [**Humanity's Last Exam**Long Phan, Alice Gatti, Ziwen Han, et al.*2025 · 235 citations*](https://md.research.elamur.ai/free/en/papers/phan-2025-humanity-s-last-exam/)
20. [**Findings of the Third Shared Task on Multimodal Machine Translation**Loïc Barrault, Fethi Bougares, Lucia Specia, et al.*2018 · 227 citations*](https://md.research.elamur.ai/free/en/papers/barrault-2018-findings-third-shared-task-multimodal/)
21. [**Generating Fact Checking Explanations**Pepa Atanasova, Jakob Grue Simonsen, Christina Lioma, et al.*2020 · 219 citations*](https://md.research.elamur.ai/free/en/papers/atanasova-2020-generating-fact-checking-explanations/)
22. [**Aya Dataset: An Open-Access Collection for Multilingual Instruction Tuning**Shivalika Singh, Freddie Vargus, Daniel D'souza, et al.*2024 · 202 citations*](https://md.research.elamur.ai/free/en/papers/singh-2024-aya-dataset-open-access-collection/)
23. [**Multilingual Part-of-Speech Tagging with Bidirectional Long Short-Term Memory Models and Auxiliary Loss**Barbara Plank, Anders Søgaard, Yoav Goldberg*2016 · 201 citations*](https://md.research.elamur.ai/free/en/papers/plank-2016-multilingual-part-speech-tagging-bidirectional/)
24. [**Multilingual Projection for Parsing Truly Low-Resource Languages**Željko Agić, Anders Johannsen, Barbara Plank, et al.*2016 · 195 citations*](https://md.research.elamur.ai/free/en/papers/agic-2016-multilingual-projection-parsing-truly-low/)
25. [**A Survey on Stance Detection for Mis- and Disinformation Identification**Momchil Hardalov, Arnav Arora, Preslav Nakov, et al.*2022 · 187 citations*](https://md.research.elamur.ai/free/en/papers/hardalov-2022-survey-stance-detection-mis-disinformation/)

[See all 76 works →](https://md.research.elamur.ai/free/en/agglomerations/copenhagen/works/)

## Works by year

### 2020s

[**2026** 1 work](https://md.research.elamur.ai/free/en/papers/zhang-2026-language-models-restoration-latin-legal/) · [**2025** 1 work](https://md.research.elamur.ai/free/en/papers/phan-2025-humanity-s-last-exam/) · [**2024** 5 works](https://md.research.elamur.ai/free/en/agglomerations/copenhagen/year/2024/) · [**2023** 9 works](https://md.research.elamur.ai/free/en/agglomerations/copenhagen/year/2023/) · [**2022** 2 works](https://md.research.elamur.ai/free/en/agglomerations/copenhagen/year/2022/) · [**2021** 10 works](https://md.research.elamur.ai/free/en/agglomerations/copenhagen/year/2021/) · [**2020** 10 works](https://md.research.elamur.ai/free/en/agglomerations/copenhagen/year/2020/)

### 2010s

[**2019** 7 works](https://md.research.elamur.ai/free/en/agglomerations/copenhagen/year/2019/) · [**2018** 5 works](https://md.research.elamur.ai/free/en/agglomerations/copenhagen/year/2018/) · [**2017** 5 works](https://md.research.elamur.ai/free/en/agglomerations/copenhagen/year/2017/) · [**2016** 14 works](https://md.research.elamur.ai/free/en/agglomerations/copenhagen/year/2016/) · [**2014** 1 work](https://md.research.elamur.ai/free/en/papers/thomas-2014-negation-scope-spelling-variation-text/) · [**2013** 2 works](https://md.research.elamur.ai/free/en/agglomerations/copenhagen/year/2013/)

### 2000s

[**2009** 1 work](https://md.research.elamur.ai/free/en/papers/jongejan-2009-automatic-training-lemmatization-rules-that/) · [**2006** 1 work](https://md.research.elamur.ai/free/en/papers/paggio-2006-annotating-information-structure-corpus-spoken/) · [**2001** 1 work](https://md.research.elamur.ai/free/en/papers/hardt-2001-generation-vp-ellipsis-corpus-based/)

### 1980s

[**1982** 1 work](https://md.research.elamur.ai/free/en/papers/maegaard-1982-transfer-finite-verb-forms-machine/)

## Concepts, shelves and topics

Of its works in our library: the number of works on each and its rating among them.

💡 Concepts 2[Lexicography · 1 work](https://md.research.elamur.ai/free/en/concepts/lexicography/) · [Part-of-speech tagging · 1 work](https://md.research.elamur.ai/free/en/concepts/part-of-speech-tagging/)

📚 Shelves 13[Natural language processing · 45 works](https://md.research.elamur.ai/free/en/shelves/nlp/) · [Linguistics & languages · 42 works](https://md.research.elamur.ai/free/en/shelves/linguistics/) · [Law · 5 works](https://md.research.elamur.ai/free/en/shelves/law/) · [AI & machine learning · 4 works](https://md.research.elamur.ai/free/en/shelves/ai/) · [English · 16 works](https://md.research.elamur.ai/free/en/shelves/en/) · [Machine translation · 11 works](https://md.research.elamur.ai/free/en/shelves/machine-translation/) · [Mathematics & statistics · 1 work](https://md.research.elamur.ai/free/en/shelves/mathematics/) · [Other · 1 work](https://md.research.elamur.ai/free/en/shelves/other/) · [Machine translation · 5 works](https://md.research.elamur.ai/free/en/shelves/mtrans/) · [Society & ethics · 1 work](https://md.research.elamur.ai/free/en/shelves/society/) · [Aramaic · 1 work](https://md.research.elamur.ai/free/en/shelves/arc/) · [History & humanities · 1 work](https://md.research.elamur.ai/free/en/shelves/history/) · [Ancient Hebrew · 1 work](https://md.research.elamur.ai/free/en/shelves/hbo/)

🧭 Topics 27[Sentiment & emotion · 10 works](https://md.research.elamur.ai/free/en/topics/sentiment-emotion/) · [Multilingual & cross-lingual NLP · 10 works](https://md.research.elamur.ai/free/en/topics/multilingual-nlp/) · [Bias, fairness & safety · 3 works](https://md.research.elamur.ai/free/en/topics/bias-fairness-safety/) · [Evaluation & benchmarks · 6 works](https://md.research.elamur.ai/free/en/topics/evaluation-benchmarks/) · [Summarization · 1 work](https://md.research.elamur.ai/free/en/topics/summarization/) · [Low-resource languages · 3 works](https://md.research.elamur.ai/free/en/topics/low-resource-languages/) · [Idioms & multiword expressions · 2 works](https://md.research.elamur.ai/free/en/topics/multiword-expressions/) · [Language models & pretraining · 3 works](https://md.research.elamur.ai/free/en/topics/language-models/) · [Legal AI · 5 works](https://md.research.elamur.ai/free/en/topics/legal-ai/) · [Interpretability · 2 works](https://md.research.elamur.ai/free/en/topics/interpretability/) · [Discourse analysis · 4 works](https://md.research.elamur.ai/free/en/topics/discourse-analysis/) · [Machine translation · 3 works](https://md.research.elamur.ai/free/en/topics/machine-translation/) · [Part-of-speech tagging · 2 works](https://md.research.elamur.ai/free/en/topics/part-of-speech-tagging/) · [Morphology · 2 works](https://md.research.elamur.ai/free/en/topics/morphology/) · [Vision & language · 2 works](https://md.research.elamur.ai/free/en/topics/multimodal/)+12 more

Agglomerations: the GHSL Urban Centre Database 2025; populations: the GHSL functional urban areas 2015 (urban centre and commuting zone), else the urban centre, as each page says (European Commission, Joint Research Centre, CC BY 4.0). An institution belongs to the urban centre its coordinates (ROR, OpenAlex) fall in, or one within 5 km. Land: Natural Earth.

- [← back to](https://md.research.elamur.ai/free/en/countries/denmark/) — Denmark

Updated 23 October 2026